Management of Isoniazid Preventive Therapy in Southern Lima, Peru: An Analysis of Health Center Characteristics

Background: Prevention of TB through the use of preventive treatment is a critical activity in the elimination of TB. Health personnel are a key element in achieving proper management of isoniazid preventive therapy (IPT) for TB contacts. This study aims to determine the association between health center characteristics and indicators of TB contact care management with isoniazid preventive therapy in southern Lima. Methods: We conducted an ecological study. Through the review of medical records, we identied children and adolescent contacts of tuberculosis patients, who initiated IPT between 2016 and 2018. We assessed bivariate associations between clinic stang and IPT initiation and completion using binomial logistic regression with robust standard errors. Results: We included 977 contacts, among whom 69% took more than a week to start IPT and 41% did not complete IPT. For those who did successfully complete IPT, 58% were not medically evaluated with three follow-up appointments. While completion of IPT and 3 clinical appointments was better among health centers with more physicians and nurses, these differences were not statistically signicant. Contact sex was associated with initiating IPT (p=0.005), and contact age was associated with completion of IPT (p=0.025) and completion of clinical evaluations (p=0.041). Conclusion: There are signicant gaps in IPT management in health centers of southern Lima, Peru, but insucient stang of health centers may not be responsible. Further research is needed to identify how IPT management can be improved, including through improving training and stang.

distribution of human resources in health centers, based on the prevalence of TB and risk of transmission in a given area.
Although the proper implementation of international standards for TB prevention could prevent the deaths of thousands of children and adolescents around the globe [7], initiation and maintenance of preventive treatment is poor [8]. Various countries have reported non-compliance with policies aimed at medical management of TB contacts [9]. Most research that focuses on TB contacts in minors less than 18 years of age evaluates the administration of the puri ed protein derived (PPD) skin test and completion of a 24-week course of isoniazid preventive therapy (IPT) [10][11][12]. There are fewer studies that evaluate other aspects of TB contact care, such as the timely onset of IPT and proper clinical evaluation with follow-up medical appointments [13]. Peruvian TB guidelines indicate that TB contacts 19 years or younger should receive a PPD test and 24 weeks of IPT if the PPD result is positive. For TB contacts less than ve years of age, the guidelines recommend IPT without requirement of a PPD test [14].
Weak health systems hinder TB control efforts [15]. The WHO classi es the health workforce as an essential building block for strong health systems. This includes a su cient, competent, and equitably distributed group of healthcare workers that can improve health outcomes [16]. Healthcare workers are especially vital in resource-constrained areas, such as Lima, where social determinants of health such as transportation, health-seeking behaviors, and material resources have been shown to impede patient access to care [17,18]. In 1990, Peru's National TB Program (NTP) became a national health priority, and since then there have been increased funds and political commitment to the NTP [19]. Despite this, however, Lima continues to face a lack of quali ed and well-trained staff in the city's most impoverished areas, where the public health system may lose healthcare staff to private or international health bodies [20].
It is essential to generate evidence about health personnel sta ng at the primary care level of decisionmaking in attempts to reduce health inequities and improve TB prevention. The objective of the study is to determine the association between health center characteristics, including available human resources, and effective IPT management in southern Lima.

Study design and setting
In 2018, we conducted an ecological study to analyze the relationship between sta ng of health centers and indicators of effective IPT management in southern Lima, Peru. The public health system in Peru is divided into regional health networks, which are managed by a Directorate of Integrated Networks (DIN) o ce. This study was conducted in the regional health network of southern Lima, under management of the Lima South DIN. In total, this DIN manages 136 health centers located in 13 city districts, covering a combined population of 1,860,382 people.
We conducted this study in 46 health centers in the Lima South DIN, located in 6 districts of Lima Metropolis (Fig. 1). These health centers were purposively sampled because they treat approximately 80% of the TB cases in the Lima South DIN and represent a range of health center levels and caseloads [21].
Each health center had a designated TB treatment program that was regulated by the NTP.

Data collection
The outcomes of interest were indicators of the quality of IPT management, assessed at the patient level. We extracted data from health records for patients meeting the following criteria: individuals 18 years of age or younger who initiated IPT and were family members or home contacts of patients receiving treatment for TB between 2016 and 2018. We excluded contacts whose information was not legible, whose start date of IPT treatment was unknown, who were diagnosed with TB after initiating IPT, and/or who suspended IPT for other medical reasons.
We evaluated the following indicators IPT management quality: 1) time between initiation of TB treatment in the index case and initiation of IPT in the contact; 2) completion status of IPT, considered complete when the contact took isoniazid for 24 weeks or 168 doses; and 3) clinical evaluation of contacts, which was considered complete when the contact attended two medical appointments during IPT and one follow-up medical appointment after IPT completion.
Health center-level predictors of interest were: 1) physician sta ng; 2) nurse sta ng; 3) number of TB cases that the health center reported in 2016; and 4) level of health center care, de ned as I-2 for an outpatient health clinic, I-3 for health centers without overnight beds, and I-4 for health centers with overnight beds. In addition, we evaluated associations between age and sex of the contacts and indicators of IPT management quality. We obtained information on physician and nurse sta ng from the Lima South DIN administrative o ce and information on contacts from their health records.

Statistical analysis
To assess the signi cance of bivariate associations between predictors and outcomes, we generated pvalues using a binomial logistic regression with robust standard errors to account for clustering at the health center level. Statistical analysis was performed using SAS v9.4. Ethical approval.
The study was approved by the Institutional Ethics Committees of the Maria Auxiliadora University (Nº 005-2019). A waiver of informed consent was granted because data were collected from routine medical records in a way that did not permit the identi cation of patients.

Results
We identi ed 1010 contacts that met the inclusion criteria and obtained information from 977 (99%). Of these, 772 individuals were contacts of index patients with sputum smear-positive TB. Table 1 describes the sociodemographic and clinical characteristics of contacts. Overall, 616 (63%) contacts were under 5 years of age, 551 (56%) were male, and 925 (95%) were contacts of index patients with pulmonary TB. Table 2 shows a description of the health centers included in the study. 41% of health centers had a fulltime physician and 13% had 2 or more nurses working in the TB service. 46% of health centers reported serving between 25-49 TB patients per year. Table 3 shows results from the bivariate analysis. While completion of IPT and 3 clinical appointments was better among health centers with more physicians and nurses, these differences were not statistically signi cant. There were also no signi cant differences in IPT management indicators based on health center caseload or level. The percentage of female contacts who had delayed IPT initiation (31%) was signi cantly larger than male contacts (24%) who delayed (p = 0.005). IPT completion was signi cantly greater (p = 0.025) in contacts age 5 and older (64%) compared to contacts under 5 years of age (56%). In addition, the percentage of contacts aged 5 and older who completed medical evaluations for IPT (46%) was statistically higher (p = 0.041) than contacts under 5 years of age (35%).

Discussion
The study found that no association between health center level, sta ng, caseload, and management of IPT in pediatric contacts of TB patients. The lack of an association between health center characteristics and IPT completion could be mediated through the role of caregivers for children, and whether children experience adverse events in treatment, as reported in other studies [22][23][24]. Services within the health center may only indirectly in uence IPT completion. In addition, because we only collected data for children who initiated IPT, we may have missed children most affected by lack of resources for IPT management -those who never initiated treatment. Other studies have shown the greatest gaps in contact management to occur during the identi cation and screening process [25,26]. These activities require trained staff, capacity to complete radiographic imaging, and supplies for PPD testingconditions that vary with the level of health center. Our ndings contrast with a previous study in Lima, Peru, which reported that contacts attending health centers with high caseloads were less likely to complete IPT [13].
We found that 72% of contacts initiated IPT at least 7 days after the index patient began TB treatment, representing a delay in IPT initiation. This is greater than a study conducted in Cuba, which reported that 23% of contacts initiated contact control late [10]. Another study in Taiwan found an association between delay in initiation of TB treatment and health system characteristics, such as living in long-term care facilities and initial visit at a primary care clinic [27]. Considering that 79% of children included in our study were contacts of patients with sputum smear-positive TB, the delay in initiation of IPT places children at greater risk for developing TB disease.
Although preventive treatment for contacts is recommended in most countries [28], completing preventive treatment with 6 months of isoniazid is a challenge. The reported completion of IPT in this study (59%) is similar to other studies conducted in Peru [11][12][13] and Ecuador [29], and greater than studies from South Africa [25] and Ethiopia [23]. The fact that IPT completion is suboptimal across a range of settings suggests that this aspect of contact management continues to be a challenge for clinical teams, particularly for nurses who are often charged with contact follow-up. One strategy that could improve treatment completion is the adoption of shorter 3 or 4-month preventive regimens that have shown better completion compared to isoniazid [30,31].
Medical evaluation of contacts is important for detecting and preventing TB disease. Our study found that the majority of contacts who completed IPT were not clinically evaluated with the recommended three follow-up appointments. This is similar to what was reported in another study in Lima that provided community support to improve TB contact management [12]. Another recent study in Peru included all contacts in their analysis, both those receiving IPT and those not, and reported only 5% completed three follow up appointments [13]. Unlike TB patients who receive directly observed therapy in clinic, contacts do not regularly attend health centers, which may represent a barrier to completing these medical appointments.
This study includes several limitations. We only collected data on contacts who completed IPT. We did not evaluate the percentage of contacts who were eligible for but did not receive IPT. This subset of contacts may have been more affected by health center resources and sta ng. Future research might evaluate the association between health center characteristics and poor prescription of IPT, a recognized gap in TB contact care in Peru [12]. Additionally, we collected clinical data from paper-based, abbreviated medical charts, and excluded charts that did not contain complete information. We attempted to improve data collection by also reviewing index case records and long-form medical records of contacts. We included health centers with higher TB caseloads, and so results reported may not apply to health centers with fewer TB cases.

Conclusions
We collected information from health centers that together report 80% of the total TB cases from the Lima South DIN. This study showed no association between health center characteristics and indicators of IPT management, suggesting that gaps in management of contacts who initiate IPT may not be due to insu cient sta ng. However, contact sex was associated with delay in initiating IPT and age was associated with completion of IPT and clinical follow-up appointments.
Health-providing institutions must promote ongoing training of health personnel to reduce gaps in contact management and subsequent TB infection in children. Further research is needed to improve the management of TB contact monitoring and management.